Total Synthesis of Petrosin, Petrosin A, and Petrosin B

Article abstract of DOI:10.1021/jo9801768

The petrosins are a family of marine alkaloids that includes the chiral, racemic isomer petrosin (1), the meso isomer petrosin A (2), and the chiral, scalemic isomer petrosin B (3). Monte Carlo molecular mechanics calculations indicated that petrosin (1) is the most stable isomer of the group, suggesting that it might be synthesized by a route that utilizes thermodynamic control for establishing the relative configurations of the eight stereocenters. The model synthesis summarized in Scheme 1 showed that intramolecular Mannich condensation is a viable route to the quinolizidone subunit of the petrosins and that this synthesis gives isomer 5, having the relative configuration found in petrosin and petrosin A, as the kinetic product. Equilibration studies with this isomer afforded an approximate equimolar mixture of 5 and diastereomer 6, having the relative configuration found in one of the two quinolizidone units of petrosin B. On the basis of this model study, a "stereo-uncontrolled" synthesis of petrosin was carried out, as summarized in Schemes 3-5. The key step of this synthesis is a "double-barrelled" intramolecular Mannich condensation of a diamino keto dialdehyde. This transformation provides crystalline petrosin in 23% yield, along with about 37% of a mixture of petrosin diastereomers. Although simple acid- and Lewis acid-mediated equilibrations of this mixture of diastereomers were not successful, the derived mixture of bis-butylimines undergoes equilibration upon treatment with protic acid to give a mixture that is greatly enriched in petrosin, relative to the other isomers. Crystallization of this equilibration mixture provided another 10% of petrosin, bringing the overall yield of petrosin to 33%. In the course of the equilibration studies, pure samples of petrosin A (2), petrosin B (3), and petrosin B′ (7) were isolated and characterized.

Full text of DOI:10.1021/jo9801768

J . Org. Chem. 1998, 63, 5001-5012
5001
Tota l Syn th esis of P etr osin , P etr osin A, a n d P etr osin B
Robert W. Scott, J ames Epperson, and Clayton H. Heathcock*
Department of Chemistry, University of California, Berkeley, California 94720
Received February 2, 1998
The petrosins are a family of marine alkaloids that includes the chiral, racemic isomer petrosin
(1), the meso isomer petrosin A (2), and the chiral, scalemic isomer petrosin B (3). Monte Carlo
molecular mechanics calculations indicated that petrosin (1) is the most stable isomer of the group,
suggesting that it might be synthesized by a route that utilizes thermodynamic control for
establishing the relative configurations of the eight stereocenters. The model synthesis summarized
in Scheme 1 showed that intramolecular Mannich condensation is a viable route to the quinolizidone
subunit of the petrosins and that this synthesis gives isomer 5, having the relative configuration
found in petrosin and petrosin A, as the kinetic product. Equilibration studies with this isomer
afforded an approximate equimolar mixture of 5 and diastereomer 6, having the relative
configuration found in one of the two quinolizidone units of petrosin B. On the basis of this model
study, a “stereo-uncontrolled” synthesis of petrosin was carried out, as summarized in Schemes
3
-5. The key step of this synthesis is a “double-barrelled” intramolecular Mannich condensation
of a diamino keto dialdehyde. This transformation provides crystalline petrosin in 23% yield, along
with about 37% of a mixture of petrosin diastereomers. Although simple acid- and Lewis acid-
mediated equilibrations of this mixture of diastereomers were not successful, the derived mixture
of bis-butylimines undergoes equilibration upon treatment with protic acid to give a mixture that
is greatly enriched in petrosin, relative to the other isomers. Crystallization of this equilibration
mixture provided another 10% of petrosin, bringing the overall yield of petrosin to 33%. In the
course of the equilibration studies, pure samples of petrosin A (2), petrosin B (3), and petrosin B′
(7) were isolated and characterized.
Between 1982 and 1988, Braekman and co-workers
reported on the isolation and structural determination
of three ichthyotoxic bis-quinolizidone alkaloids from the
sponge Petrosia seriata, collected from the waters off of
Papua New Guinea.1
,2
The three diastereomers are
petrosin (1), petrosin A (2), and petrosin B (3). These
natural products present a fascinating stereochemical
and biosynthetic puzzle. Although it contains eight
stereocenters, petrosin (1) is racemic as isolated. The
relative configuration of petrosin was firmly established
by X-ray crystallography. Petrosin A (2) is also devoid
of optical activity, and the presence of only 15 peaks in
the C NMR spectrum for this isomer confirmed a
symmetrical system. A 2-D NMR study of petrosin and
petrosin A established that the quinolizidone units in
these compounds have the same relative configuration.
This led to the meso structure 2, in which the two halves
of the molecule are enantiomeric. Petrosin B (3), on the
other hand, is optically active ([R]579 ) -12). In this
isomer, one of the quinolizidone units has the same
relative configuration as that found in petrosin and
petrosin A, but the other quinolizidone has a different
relative configuration. The structure was determined by
1
3
the sponge Xestospongia sp.3 In this work, the racemic
nature of petrosin and the meso structure of petrosin A
were confirmed by reduction of each isomer to the
diequatorial diol, formation of the diester with a chiral
carboxylic acid, and examination of the NMR spectra of
these products. These experiments provided unambigu-
ous proof that petrosin is racemic, rather than simply
having a very low optical rotation.
Since it is uncommon for natural products to be
biosynthesized in racemic form, we wondered if petrosin
and petrosin A might be products of some post-biosyn-
thetic equilibration. We evaluated this possibility by
carrying out a series of molecular mechanics calculations.
Minimizations were first carried out for the three chair-
chair conformations of each of the eight diastereomeric
forms of trimethylquinolizidone 4.4 Two low-energy
structures of approximately equal energy, 5 and 6, were
2
-D NMR analysis and comparison to the previously
obtained data for 1 and 2.
Petrosin and petrosin A have also been isolated from
(
1) (a) Braekman, J . C.; Daloze, D.; Macedo de Abreu, P.; Piccini-
Leopardi, C.; Germain, G.; Van Meerssche, M. Tetrahedron Lett. 1982,
3, 4277. (b) Braekman, J . C.; Daloze, D.; Defay, N.; Zimmerman, D.
2
Bull. Soc. Chim. Belg. 1984, 93, 941. (c) Braekman, J . C.; Daloze, D.;
Cimino, G.; Trivellone, E. Bull Soc. Chim Belg. 1988, 97, 519.
(3) Kobayashi, M.; Kawazoe, K.; Kitagawa, I. Tetrahedron Lett.
1989, 30, 4149.
(2) For an excellent recent review of this structurally unique class
of marine natural products, see: Matzanke, N.; Gregg, R. J .; Weinreb,
S. M. Org. Prep. Proc. Int. 1998, 1-51.
(4) Calculated molecular mechanics energies of these 24 structures
are included in the Supporting Information.
S0022-3263(98)00176-5 CCC: $15.00 © 1998 American Chemical Society
Published on Web 06/26/1998

5
002 J . Org. Chem., Vol. 63, No. 15, 1998
Scott et al.
Sch em e 1a
Ta ble 1. Str u ctu r es a n d Ca lcu la ted Rela tive En er gies
-
1
(
k ca l m ol ) of P etr osin Dia ster eom er s
a
Key: (a) (CH2OH)2, p-TsOH; (b) LiAlH4, THF; (c) 5% HOAc,
1
00 °C; (d) BF3, CH2Cl2.
6
to enone 13 to obtain 14 as a diastereomeric mixture.
Treatment of 14 with aqueous acetic acid provided
bicyclic amino ketone 5 in 70% yield, accompanied by
about 12% of amino ketone 6 and smaller amounts of
several other unidentified isomers. The major isomer
was isolated as a crystalline picrate and characterized
by single-crystal X-ray analysis. The minor isomer was
found. Isomer 5 represents the relative configuration
present in petrosin, petrosin A, and one-half of petrosin
B, whereas isomer 6 contains the unique configuration
present in the novel half of petrosin B.
1
identified as 6 by H NMR analysis. Treatment of a 5.4:1
mixture of 5 and 6 with boron trifluoride in methylene
chloride gave an approximate equimolar mixture of the
two diastereomers.
The equilibration experiments with 5 and 6 are in
excellent agreement with the molecular mechanics cal-
culations, which showed these two isomers to be of
approximately equal energy and at least 1.5 kcal/mol
more stable than any other diastereomeric trimeth-
ylquinolizidone. It is also notable that the kinetic product
of the intramolecular Mannich reaction, isomer 5, has
the relative configuration found in both quinolizidone
subunits of the two most abundant petrosin stereoiso-
mers, petrosin and petrosin A. This kinetic preference
is understandable if one considers the transition states
for the C-C bond-forming step in the two Mannich
reactions (Scheme 2). In the transition state leading to
isomer 5, the enol double bond attacks the immonium
ion trans to the ring methyl group, leading initially to a
“cis-decalinic” quinolidine, which isomerizes to 5 by
inversion of the tertiary amine and epimerization of one
of the methyl stereocenters adjacent to the carbonyl
group. In the isomeric transition state leading to 6, the
enol must attack the immonium ion cis to the ring methyl
group, resulting in the indicated nonbonded interaction.
The experimental result of the model synthesis sum-
marized in Scheme 1 is consistent with a small difference
of about 0.6 kcal/mol in the energies of these two
transition states (ratio of 5:6 ) 6:1). Control experiments
showed that 5 and 6 do not equilibrate under the
conditions of the Mannich reaction in which they are
formed.
A series of Monte Carlo global minimizations were then
carried out using various combinations of the two enan-
tiomeric forms of 5 and 6, randomizing the cyclohexade-
cane conformations. In this way, low-energy conforma-
tions of six diastereomeric forms of petrosin were
generated. These six diastereomers correspond to the
three known natural products, petrosin (1), petrosin A
(
2), petrosin B (3), and three unknown diastereomers,
5
petrosin B′ (7), petrosin C (8), and petrosin D (9). The
calculated low-energy conformations of each of these
diastereomers is indicated along with their structures in
Table 1.
The computed low-energy conformation of petrosin is
1
identical to the X-ray structure and lower in energy than
the low-energy conformations of petrosins A, B, B′, C, and
D. These calculations suggested the interesting possibil-
ity of a synthesis based on thermodynamic control of
relative configuration at the eight stereocenters. To this
end, we carried out the simple model synthesis depicted
in Scheme 1. Cyanoethylation of propanal, by way of the
piperidine enamine, provided cyano aldehyde 10, which
was converted into acetal 11 by treatment with ethylene
glycol. Reduction of the cyano group with lithium
aluminum hydride provided amine 12, which was added
At this point, it became clear that it should be possible
to synthesize petrosin by a route involving a double-
(
5) Prior work did not actually distinguish between structures 3 and
(6) Gore, W. P.; Pearce, G. T.; Silverstein, R. M. J . Org. Chem. 1975,
40, 170.
7
for petrosin B.

Synthesis of Petrosin, Petrosin A, and Petrosin B
J . Org. Chem., Vol. 63, No. 15, 1998 5003
Sch em e 2
Sch em e 3a
a
Key: (a) (i) O3; (ii) Zn, HOAc; (b) (i) pyrrolidine, K2CO3, (ii) CH2)CHCN, (iii) H2O; (c) NaBH4; (d) TBS-Cl; (e) DIBAL, -95 °C; (f)
CH3CH)CO2Li2; (g) CH2N2; (h) H2, PtO2, EtOAc, HOAc.
barrelled intramolecular Mannich reaction, without pay-
ing any particular attention to stereocontrol.7,8 The
Mannich reactions leading to the quinolizidone subunits
should give mainly “type 5” relative stereochemistry for
kinetic reasons. Thus, of the six petrosin isomers de-
picted in Table 1, a synthesis based on intramolecular
Mannich reactions should yield mainly petrosin and
petrosin A, smaller amounts of petrosins B and B′, and
relatively minor amounts of petrosins C and D. If we
were then able to achieve equilibration of this kinetic
mixture (for example, by some acid-mediated method
involving retro-Mannich reactions), we might be able to
enrich the petrosin-petrosin A mixture significantly in
favor of petrosin. With this object in mind, we set about
a “stereo-uncontrolled” synthesis of petrosin.
Our synthesis began with the natural fatty-acid de-
rivative methyl oleate, which was ozonized to obtain
methyl azelaldehyde (15) (Scheme 3). The aldehyde was
cyanoethylated by way of its pyrrolidine enamine to
obtain 16.9 The aldehyde was reduced to obtain alcohol
1
7, which was protected as the tert-butyldimethylsilyl
(
7) For a preliminary account of this work, see: Scott, R. W.;
Epperson, J .; Heathcock, C. H. J . Am. Chem. Soc. 1994, 116, 8853.
8) Hoye and co-workers have reported a thermodynamically con-
ether 18. This material was reduced to cyano aldehyde
1
tently obtained by this method with yields in the mid-
9 with DIBAL-H at -95 °C. Compound 19 was consis-
(
trolled synthesis of the biosynthetically related compound xestospongin
A: Hoye, T. R.; North, J . T.; Yao, L. J . J . Am. Chem. Soc. 1994, 116,
2
1
617. Hoye, T. R.; Ye, Z.; Yao, L. J .; North, J . T. J . Am. Chem. Soc.
996, 118, 12074.
(9) Stork, G. A.; Brizzolara, H.; Landesman, H.; Szmuszkovicz, J .;
Terrell, R. J . Am. Chem. Soc. 1963, 85, 207.

5
004 J . Org. Chem., Vol. 63, No. 15, 1998
Scott et al.
Ta ble 2. Cou p lin g of Acid 20 w ith Am in e 21
high yield with only trace removal of the silyl group. The
large amount of catalyst was required to speed the
reaction enough that it was complete before significant
desilylation occurs.
catalyst in
entry reduction of 20
coupling
reagents
a
solvent yield of 22, %
1
2
3
4
5
6
Raney Ni
Raney Ni
Raney Ni
Raney Ni
Raney Ni
PtO2, HOAc
DPPA
CDI
EDC, HOBT DMF
DCC, PFP
DCC, HOBT THF
DCC, HOBT THF
CH2Cl2
CH2Cl2
20-33
21
40-48
49
46-48
70
b
CH2Cl2
a
DPPA ) diphenyl phosphorazidate; CDI ) carbonyldiimida-
zole; EDC ) 1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide hy-
drochloride; DCC ) dicyclohexylcarbodiimide; HOBT ) N-hydroxy-
b
benzotriazole; PFP ) pentafluorophenol. In this experiment, the
pentafluorophenyl ester was preformed and the amine was then
added.
8
0% range, with the balance of the material being mainly
the cyano alcohol resulting from further reduction of the
aldehyde. Maintaining a low temperature in this reac-
tion is crucial, and even careful reaction at -78 °C
proceeded with 10-15% more overreduction to the alco-
hol. Treatment of aldehyde 19 with the dianion of
The optimized procedure for preparation of amine 21
resulted in a good yield of amide 22 when coupling was
carried out with dicyclohexylcarbodiimide and N-hy-
droxybenzotriazole.11 Our first attempts to achieve mac-
rocyclization were carried out directly on the amino acid
with Mukiayama’s reagent.12 These reactions led to
several products, none of which could be identified as the
desired macrolactam. In our next approach, compound
propionic acid gave acid 20 as a 1:1 mixture of syn and
anti diastereomers.10 At this point, the material was
partitioned, as this precursor is used for both halves of
the molecule. One portion is treated with diazomethane
to produce the methyl ester, followed by reduction of the
nitrile to amine 21. This provides us with two differen-
tially functionalized pieces to proceed with the sequential
amide bond formations, as one segment has a free
carboxylic acid and a latent amine (nitrile), whereas the
other contains a protected carboxylic acid (ester) and free
amine.
2
2 was treated with hydrogen over Adams’ catalyst in
ethanolic hydrochloric acid. In this way, the cyano group
was reduced to a primary amine, which was trapped as
the hydrochloride, and both silyl protecting groups were
removed. After neutralization with NaHCO
was protected as the BOC derivative to give 23 (Scheme
). This compound possesses suitable functionality for
3
, the amine
Reduction of the methyl ester of 20 was initially carried
out over Ra-Ni catalyst under basic conditions, which
produced the amino ester 21 in high yield. However, the
primary amine was difficult to purify; therefore, 21 was
used crude from the hydrogenation reaction in the amide
coupling reaction with acid 20. Under several different
amide formation conditions, amide 22 was obtained in
no greater than 50% yield (Table 2, entries 1-5). Ad-
ditionally, it was discovered that acid 20 couples cleanly
with a model pure amine and that amine 21 suffered the
same low yields when attempts to couple it to a pure
model acid were tried. Although the amine from the Ra-
Ni reduction was clean by NMR, it was known that it
contained some amount of nickel, as the viscous oil was
a slightly green color and the recovery yield always
several percent over quantitative. It is believed that this
nickel complexed to the amine somehow inhibits the
amide formation reaction, leading to the lower than
expected yields of product. However, this realization
presented a new problem, as the other common way to
deal with the amine reduction is to perform the hydro-
genation with typical catalysts (Pd or Pt) in the presence
of an acid source to protonate the amine as formed,
thereby limiting its ability to poison the catalyst. Since
our substrate contains an acid-labile TBS protecting
group; not surprisingly, most hydrogenations in the
presence of acid led to the removal of this group.
Eventually through trial and error, it was discovered that
by carrying out the reduction in EtOAc with approxi-
mately 8-9 equiv of acetic acid over a large amount of
4
a stepwise activation/macrolactamization procedure, with-
out the necessity of protecting the hydroxy groups. The
successful macrolactamization was accomplished by a
four-step procedure that was optimized on the basis of
several literature precedents.13 The methyl ester of 23
was saponified and the resulting carboxylic acid con-
verted to the pentafluorophenyl ester by treatment with
dicyclohexylcarbodiimide and pentafluorophenol. Depro-
2 2
tection of the BOC group with standard TFA/CH Cl
conditions followed by concentration of the crude reaction
mixture did lead to successful deprotection; however, as
a side reaction, varying amounts of trifluoroacetates were
1
formed on the free hydroxyls (as determined by H and
1
9
F NMR and FAB-MS). To avoid this undesired reac-
tion, the crude active ester was treated with anhydrous
HCl in dioxane to remove the BOC group, and simulta-
neously protect the amine as the hydrochloride salt. A
solution of this hydrochloride salt in dioxane-THF was
slowly added with a syringe pump to a hot 5:1 dioxane-
pyridine solution, providing the desired 28-membered
macrolactam 24 in 78% overall yield for the three steps.
The amide linkages were reduced with lithium alumi-
num hydride to the diamine tetraol. Attempted oxidation
(
11) Windridge, G. C.; J orgensen, E. C. J . Am. Chem. Soc. 1971,
9
3, 6318.
(12) Bald, E.; Saigo, K.; Mukaiyama, T. Chem. Lett. 1975, 1163.
(
13) (a) Evans, D. A.; Ellman, J . A. J . Am. Chem. Soc. 1989, 111,
1
063. (b) Schmidt, U.; Meyer, R.; Leitenberger, V.; Griesser, H.;
Lieberknecht, A. Synthesis 1982, 1025. (c) Schmidt, U.; Leitenberger,
V.; Griesser, H.; Schmidt, J .; Meyer, R. Synthesis 1992, 1248. (d)
Schmidt, U.; Griesser, H. Angew. Chem., Int. Ed. Engl. 1981, 20, 280.
PtO
2
catalyst (∼30 mol %), the nitrile could be reduced
under 3 atm of H
2
to the amine hydroacetate salt in very
(e) Schmidt, U.; Kroner, M.; Griesser, H. Synthesis 1991, 294. (f)
Schmidt, U.; Utz, R.; Lieberknecht, A.; Griesser, H.; Potzolli, B.;
Wagner, K.; Fischer, P. Synthesis 1987, 236.
(10) Moersch, G. W.; Burkett, A. R. J . Org. Chem. 1971, 36, 1149.

Synthesis of Petrosin, Petrosin A, and Petrosin B
J . Org. Chem., Vol. 63, No. 15, 1998 5005
Sch em e 4a
F igu r e 1. Portion of the 13C NMR spectrum of a typical
cyclization mixture (Scheme 5).
resonance for the C10 methine is well-removed from
other resonances in petrosin and its isomers. Further-
more, this carbon is believed to be in a similar environ-
ment for all the isomers, attached to a nitrogen and
flanked by one equatorial and one axial side-chain
substituent, so the NOE effects and relaxation delays of
this carbon should be similar enough in all the isomers
of interest that 13C NMR line heights can be used as an
excellent measure of isomer ratios. An example of the
7
0-72 ppm region of the 13C NMR spectrum for a typical
cyclization mixture, prior to crystallization of the petrosin
fraction, is shown in Figure 1. At the outset, we had the
published chemical shifts for the C-10 methine reso-
nances in the three isomeric natural products, petrosin
a
Key: (a) (i) 0.23 M HCl, EtOH, H2, PtO2; (ii) (Boc)2O, dioxane,
H2O; (b) 1 M NaOH, MeOH, THF; (c) DCC, C6F5OH; (d) (i) 6 N
HCl, dioxane, (ii) syringe pump into dioxane/pyridine.
(1), petrosin A (2), and petrosin B (3). In the course of
this study, we were able to isolate pure samples of each
of these isomers, in addition to petrosin B′ (7) (vide infra).
Evaluation of the C NMR peak heights in Figure 1 gives
of the diamine-tetraol with (COCl)
2
/DMSO, SO
3
•pyridine,
13
CrO •pyridine, AgCO /Celite, Na Cr O
3
3
2
2 7
/DMSO, or Dess-
a kinetic isomer ratio of 48:22:17:13 for these four
isomers. In addition, we subsequently acquired synthetic
Martin periodinane led to unidentifiable products in all
cases. These reactions were monitored by TLC and H
NMR using authentic natural material as a reference
sample and it can be concluded that no more than trace
amounts of petrosin isomers were formed under these
conditions. As a result, the crude diamine obtained from
1
15
samples of petrosins C and D (8 and 9), permitting us
to identify small amounts of these isomers in the kinetic
cyclization mixture.
With access to reasonable amounts of petrosin and the
isomeric mixture, we commenced a series of equilibration
the LiAlH
4
reduction was protected as the bis-BOC
13
experiments, using C NMR spectroscopy as the princi-
derivative, which seemed to eliminate the competing
reactions in the oxidation as well as provide a product
that could be purified after the LiAlH reduction. The
4
hydroxy groups of the di-BOC-protected compound were
oxidized with Dess-Martin periodinane,14 which ac-
complishes the tetraoxidation quickly (<30 min) and
efficiently. The BOC groups were removed with HCl in
aqueous ethanol, and the intramolecular Mannich con-
densation was achieved by treatment of the crude product
with 0.2 M acetic acid in relfuxing ethanol.
pal analytical tool. Several attempts to equilibrate the
mixture of petrosin isomers remaining after crystalliza-
tion of most of the petrosin led only to recovered mixtures
that appeared by NMR to be largely unchanged. The
most definitive of this series of experiments were carried
out with BF
conditions that successfully equilibrated quinolizidones
and 6. However, these experiments gave unchanged
3
2
•Et O under the same or more forcing
5
starting material or resulted in general decomposition
under the more forcing conditions. To eliminate the
possibility of reagent or concentration differences, we
carried out careful comparison experiments with quino-
lizidones 5 and 6, and their equilibration was shown to
be quite reproducible. Since the steric environment in
the petrosin skeleton seems modeled by 5 and 6, we
attribute the difficulty of the equilibration to the dimeric
nature of the petrosin skeleton. As shown in Scheme 6,
to equilibrate all stereocenters within one quinolizidone
unit, it is only necessary to enolize the ketone and do
retro-Mannich and Mannich reactions. However, to fully
The cyclization product, obtained in approximately 60%
for the three-step procedure, was a mixture of petrosin
and several of its stereoisomers. Crystalline petrosin,
isolated from the crude product in 23% yield, was found
1
13
to be identical by H NMR, C NMR, TLC, and melting
point with an authentic sample provided by Professor
2
Kitagawa. The mixture of isomers can be analyzed by
1
3
C NMR, using the resonance for the methine carbon
attached to nitrogen at about 70-72 ppm. This region
of the 13C NMR spectrum is relatively uncluttered, as the
(
14) (a) Dess, D. B.; Martin, J . C. J . Am. Chem. Soc. 1991, 113, 7277.
(15) Brown, R. C. D.; Norman, T.; Heathcock, C. H J . Org. Chem.
1998, 63, 0000.
(
b) Ireland, R. E.; Liu, L. J . Org. Chem. 1993, 58, 2899.

5
006 J . Org. Chem., Vol. 63, No. 15, 1998
Scott et al.
Sch em e 5a
stereocenter is unchanged. On the other hand, to inter-
convert petrosin (1) with its diastereomer petrosin A (2),
it is necessary that the intermediate immonium ions 32
or 34 be deprotonated to enamine 33, since the overall
transformation requires epimerization of one of the C7
stereocenters. We think that the low probability of the
deprotonation step is one factor that makes it difficult
to equilibrate the petrosins under the same conditions
that suffice to equilibrate 5 and 6.
Of course, there is another major difference between
5
and 1, in that compound 5 has only one nitrogen,
whereas petrosin and its isomers have two. In order for
the retro-Mannich reaction to occur, it is necessary that
the carbonyl oxygen be coordinated with the acid while
the nitrogen is not. Because of the large difference in
3
2
basicity of an sp nitrogen and an sp oxygen, this is a
low-probability event even with 5. If we envision even
one of the nitrogens being associated with the Lewis acid,
it could greatly decrease the basicity of the oxygen lone
pair, even on the other end of the macrocycle, as the
molecule itself would already carry a positive charge. The
combination of these events may prevent the retro-
Mannich reaction for the dimeric petrosin case.
We thought that we might ameloriate the difficulties
by carrying out the equilibrations on the bis-imine
derivatives, rather than on the ketones. To this end, the
mixture of petrosin isomers was dissolved in butylamine
and heated at 85-90 °C for 1-2 days in the presence of
molecular sieves (Scheme 7). This treatment did convert
the mixture into a mixture of bis-imines, as determined
by IR spectroscopy of the crude reaction mixture. The
crude bis-imine was then subjected to various acidic
conditions in order to effect a retro-Mannich equilibration
pathway.16 Several different conditions were screened,
with the ratios of the known petrosin isomers compared
a
Key: LiAlH4; (b) Boc2O; (c) Dess-Martin; (d) 1 M HCl; (e) 0.2
M HOAc.
equilibrate one quinolizidone system relative to the other,
one must deprotonate the intermediate immonium ion
to the corresponding enamine. That is, trimethylquino-
lizidone 5 can be interconverted with its diastereomer 6
by undergoing a retro-Mannich reaction to immonium ion
8, which can then close on the other face to give 29.
After epimerization of the C3-methyl stereocenter, dia-
stereomer 6 is obtained. In this transformation, the C7
1
3
by line height of the C-10 resonance in the C NMR
spectrum. The results of some of these experiments are
presented in Table 3, and the partial 13C NMR spectrum
of a typical run is shown in Figure 2.
2
Sch em e 6

Synthesis of Petrosin, Petrosin A, and Petrosin B
J . Org. Chem., Vol. 63, No. 15, 1998 5007
Ta ble 3. Equ ilibr a tion of P etr osin Bis-Im in es (Sch em e 7)
a
a
entry
acid
solvent
time (h)
T (°C)
initial ratio 1:2:3:7
final ratio 1:2:3:7
1
2
3
4
5
6
7
PrNH3 OAc
Et3N, Et3N•HCl
PrNH3 OAc
Et3N, Et3N•HCl
pyrrolidine•HOAc
PrNH3 OAc
dichloroethane
dichloroethane
dichloroethane
DMF
pyrrolidine
dichloroethane
dichloroethane
108
25
157
67
60
50
90
90
90
95
95
90
95
36:39:18:6
19:52:21:8
22:38:26:14
18:46:24:13
17:45:26:13
100:0:0:0
54:25:13:8
36:40:15:9
50:21:18:11
54:32:10:5
46:29:15:10
90:0:10:0
Et3N, Et3N•HCl
72
100:0:0:0
84:3:13:0
a
Ratios are given for petrosin (1), petrosin A (2), petrosin B (3), and petrosin B′ (7), normalized to a total of 100%. Small amounts of
petrosins C and D (8 and 9) and other unidentified compounds are also present.
Sch em e 7
to equilibrate three stereocenters (marked with X in
Scheme 8); the retro-Mannich reaction must occur, but
it is not necessary that the intermediate immonium ion
be deprotonated to the corresponding enamine. To
convert 1 (or 3) to petrosin A (2), equilibration of the
position marked Y must also occur.
It is interesting to note that the bis-imine equilibration
provides clear enrichment in petrosin in each experiment.
However, the ratio of petrosin B to petrosin B′ is not
changed very much, even though the transformation that
converts petrosin A to petrosin would also convert
petrosin B′ to petrosin B. However, it should be remem-
bered that these ratios are obtained by simple measure-
ment of 13C NMR peak heights and are not very accurate
for the minor isomers.
In light of the equilibration of the petrosin skeleton
isomers to produce mainly petrosin, we thought we might
improve the synthesis by adding an equilibration step
after crystallization of petrosin from the kinetic cycliza-
tion mixture (Scheme 5). Though the equilibration
experiments described always proceed with some decom-
position, even one cycle does significantly improve the
yield of petrosin. Specifically, the final sequence pro-
duces a 62% yield of isomers, approximately 1/3 of which
is pure petrosin obtained by crystallization. Formation
of the bis-imine of the isomer mixture remaining after
petrosin removal followed by equilibration with propyl-
ammonium acetate in 1,2-dichloroethane leads, after
aqueous workup, to an 80% recovery of material, which
is now a 1:2 mixture of petrosin/other petrosin isomers.
In this mixture, petrosin is the major single isomer
present. Crystallization of petrosin from this mixture
provides an additional 10% of this isomer. This equili-
bration process raises the total yield of crystalline
petrosin for the final sequence to 33%, while the com-
bined yield of petrosin isomers (mostly petrosin A and
petrosin B) is reduced to 20%.
The data in Table 3 clearly show that equilibration of
the bis-imine derivative leads to mixtures that are
enriched in petrosin, relative to the starting sample.
However, even with this more favorable system, we are
far from equilibrium, as shown by entries 6 and 7 in
Table 2, which are two runs starting with pure, crystal-
line petrosin. The pure petrosin equilibration experi-
ments are important for several reasons. First, they
represent the “reverse” of the equilibrium proposed in
the isomer mixture experiments and provide further
evidence that these isomers are interconverted. This is
important as it eliminates the concern that the unchar-
acterized “isomers” in our mixture are really some other
species that we have misidentified and that produce
petrosin by some other pathway under the “equilibration”
conditions. Second, entries 6 and 7 show that the first
isomer formed by equilibration is petrosin B. As has been
previously discussed, and is illustrated in Scheme 6, it
is easier to interconvert the quinolizidone centers relative
to one another than to equilibrate C7 relative to C7′.
Therefore, the observation that petrosin gave rise only
to petrosin B in the experiment shown in entry 6 is
consistent with these two isomers having the same
relative configuration at C7 and C7′ (Scheme 8). That
is, to convert petrosin (1) to structure 3, it is necessary
In the course of studying the equilibration of petrosin
and its isomers, we were able to obtain reasonably pure
samples of the three known isomers, petrosin (1), petrosin
A (2), and petrosin B (3), as well as the previously
unknown isomer petrosin B′ (7). Petrosin is the easiest
isomer to obtain in a pure state, as it crystallizes directly.
The other isomers were isolated by repeated chromatog-
raphies of mixtures of the isomers or of the diols produced
by NaBH
4
. The detailed separation scheme is given in
1
3
the Supporting Information. The C NMR spectrum of
petrosin B′ is quite similar to that of the known petrosin
B, as is true with petrosin and petrosin A, which also
differ only in the absolute stereochemical relationship of
the two quinolizidone units. In particular, the spectra
of petrosin and petrosin A, which are both symmetrical,
are much simpler than those of petrosin B and petrosin
B′, which have two diastereomeric quinolizidone units.
(16) A control experiment where the bis-imine was formed and
subsequently hydrolyzed showed that negligible equilibration occurs
during this step.

5
008 J . Org. Chem., Vol. 63, No. 15, 1998
Scott et al.
F igu r e 2. Portion of the 13C NMR spectrum of a typical equilibration mixture (Scheme 7).
1
F igu r e 3. Portions of the H NMR spectra of petrosin (1), petrosin A (2), petrosin (B) (3), and petrosin B′ (7).
1
Sch em e 8a
The H NMR spectrum of petrosin B and petrosin B′ both
show a characteristic “triplet” structure at about 2.4 ppm,
due to the indicated axial proton, which has large
coupling constants with the axial angular hydrogen and
one of the hydrogens of the linking pentamethylene
chain. These key regions of the spectra are shown in
Figure 3.
In summary, we have worked out a relatively straight-
forward synthesis of petrosin that totally ignores the
issue of stereochemistry in its design. The final “double
Mannich cyclization” provides crystalline petrosin (1) in
more than 30% yield. In addition, we have isolated the
other two diastereomeric natural products, petrosin A (2)
and petrosin B (3), in pure form. Finally, we have
isolated and characterized the previously unknown dia-
stereomer, petrosin B′ (7), and obtained evidence that
permits us to assign its relative configuration. In the
following paper in this issue,15 we describe the synthesis
and characterization of two other petrosin diastereomers,
petrosin C (8) and petrosin D (9).
a
For clarity and simplicity of presentation, the enantiomer
depicted for petrosin has opposite handedness to that depicted in
the remainder of the paper.
perhaps probable, that the petrosin isomers would in-
terconvert quickly under mild acid catalysis, thereby
producing a mixture of isomers, the composition of which
is governed by thermodynamic stability of the various
isomers. However, this theory seems unlikely if one
At the onset of the project, we thought it was possible,

Synthesis of Petrosin, Petrosin A, and Petrosin B
J . Org. Chem., Vol. 63, No. 15, 1998 5009
Sch em e 9
tetrahydrofuran (THF) were distilled under nitrogen from
sodium/benzophenone immediately prior to use. Methylene
2 2
chloride (CH Cl ), pyrrolidine, diisopropylamine, and triethyl-
amine were distilled under nitrogen from calcium hydride prior
to use. Reagents were used as received from commercial
suppliers unless otherwise noted. The concentration of com-
mercially available n-butyllithium in hexanes was periodically
checked by titration of diphenylacetic acid. Unless otherwise
specified, extracts were dried with anhydrous magnesium
sulfate and concentrated under reduced pressure with a rotary
evaporator. Unless otherwise indicated, IR spectra were of
thin films on NaCl plates and NMR spectra measured in
CDCl
δ scale using CHCl
constants reported in Hz.
-F or m ylva ler on itr ile Dioxola n e (11). A solution of
3
. The NMR chemical shifts are reported in ppm of the
3
as an internal reference and the coupling
4
considers the drastically differing ratios of the isomers
that were found in the isolations by Braekman (ca. 7:1
ratio of petrosin/petrosin A) and Kitagawa (ca. 1:1
mixture of petrosin/petrosin A). In addition, whereas
Braekman found a small amount of petrosin B in his
isolation, the Kitagawa isolation yielded none of this
isomer. One possible explanation for this discrepancy
may be the labor of isolation of the natural products from
the naturally occurring mixtures. The repeated and
painstaking separations necessary to obtain pure com-
pounds undoubtedly lead to isolation ratios that do not
reflect the naturally occurring ratios but rather reflect
the ability to obtain pure fractions chromatographically
or by crystallization. However, it must be remembered
that petrosin A and petrosin B are practically insepa-
rable, and it is therefore unlikely that a significant
quantity of petrosin B existed in the Xestospongia sp.
isolation of Kitagawa because it would have coeluted with
petrosin A. Furthermore, since we have been unable to
find simple acidic conditions that will bring about equili-
bration of the petrosin isomers, it is highly unlikely that
petrosin is racemic because of equilibration during isola-
tion and laboratory processing.
4-formylvaleronitrile (4.64 g, 41.8 mmol), ethylene glycol (10.0
g, 161 mmol), p-toluenesulfonic acid (0.4 g, 2.1 mmol), and
benzene (100 mL) was refluxed for 16 h with azeotropic
removal of water. The solution was allowed to cool to room
temperature, washed with saturated solutions of NaCl, NaH-
3
CO , and NaCl again, dried, and concentrated. The resulting
oil was purified by distillation from a Kugelrohr apparatus to
yield 5.57 g (86%) of 11 (bp 75-77 °C, 2.5 Torr) as a clear oil.
-
1
1
IR: 2260 cm
(
3
2
.
H NMR: δ 0.98 (d, 3, J ) 6.8), 1.52-1.56
m, 1), 1.83-1.89 (m, 2), 2.15-2.47 (m, 2), 3.83-3.87 (m, 2),
.91-3.94 (m, 2), 4.66 (d, 1, J ) 4.1). 13C NMR: δ 14.12, 15.17,
7.11, 35.89, 64.88, 65.02, 106.85, 119.80. Anal. Calcd for
8 2
C H13NO : C, 61.91; H, 8.44; N, 9.03. Found: C, 62.0; H, 8.19;
N, 9.07.
-F or m ylp en tyla m in e Dioxola n e (12). Lithium alumi-
4
num hydride (1.94 g, 51.1 mmol) was suspended in THF (80
mL), and a solution of 11 (6.47, 41.7 mmol) in THF (5 mL)
was added slowly from a dropping funnel. The reaction
mixture was stirred for 0.5 h at room temperature and then
quenched by addition of 2.0 mL of H
NaOH, and 6.0 mL of H O. The solution was filtered, washed
2
O, 2.0 mL of 15% aqueous
2
with saturated NaCl solution, and concentrated. The resulting
oil was purified by distillation from a Kugelrohr apparatus to
yield 5.270 g (80%) of 12 (bp 118-120 °C, 25 Torr) as a clear
-
1
1
oil. IR: 3380, 3300 cm
.
H NMR: δ 0.95 (d, 3, J ) 6.8),
1.15-1.22 (m, 3), 1.42-1.44 (m, 1), 1.54-1.58 (m, 2), 1.69-
1
2
6
.72 (m, 1), 2.67-2.70 (m, 2), 3.83-3.87 (m, 2), 3.93-3.97 (m,
However, the successful bis-imine equilibrations may
provide a clue to the mystery of why petrosin is racemic
whereas petrosin B is optically active. It may be that
the petrosins are all biosynthesized in optically active
form but that there is an in vivo analogue of our bis-
imine equilibration that permits the retro-Mannich se-
quence to operate, thereby equilibrating petrosin with its
), 5.32 (d, 1, J ) 4.5). 13C NMR: δ 28.66, 31.43, 36.77, 42.50,
8 2
4.96, 107.56, 110.47. Anal. Calcd for C H17NO : C, 60.35;
H, 10.76. Found: C, 60.59; H, 10.52.
Qu in olizid on es 5 a n d 6. Amine 12 (439 mg, 2.76 mmol)
was dissolved in MeOH (3.0 mL) and cooled to 0 °C. Enone
13 (275 mg, 2.80 mmol) dissolved in MeOH (1.0 mL) was slowly
added from an addition funnel. The solution was allowed to
warm to room temperature, stirred for 4 h, and concentrated
and the residue taken up in 5% AcOH and heated to 100 °C
for 4 h. The cooled reaction mixture was quenched with a 5%
meso diastereomer, petrosin A. The crucial point would
then be that, because of A1,3 strain, the imine can only
easily form if the quinolizidone has the petrosin-type
relative configuration such as that found in compound 5
3 2 2
solution of NaHCO and extracted into CH Cl . The solution
was dried and concentrated to afford 510 mg (95%) of a brown
(Scheme 9). Thus, the bis-imines of petrosin and petrosin
oil, which was a mixture of isomers (mainly 5 and 6 in a ratio
A could equilibrate with one another and since petrosin
A is a meso compound, the isolated petrosin would be
racemic. On the other hand, petrosin B could only
equilibrate with its optically active diastereomer petrosin
B′. Extension of this theory to the currently unknown
diastereomers petrosin C and petrosin D suggests that
these isomers would not be able to equilibrate with one
another. Therefore, if petrosin C is eventually found as
a natural product, it should be optically active.
of approximately 5.4:1, identified by 1H NMR and GC/MS).
Isomer 5 was isolated in approximately 70% yield by MPLC.
-
1
1
IR: 2805, 2760, 1720 cm
.
H NMR: δ 0.82 (d, 3, J ) 6.5),
0
1
.96 (d, 3, J ) 6.5), 1.05 (m, 1), 1.17 (d, 3, J ) 7.2), 1.85 (m, 4),
.87 (m, 1), 1.95 (m, 2), 2.58 (m, 1), 2.93 (m, 2), 3.02 (m, 1).
C NMR: δ 11.2, 12.0, 17.6, 25.2, 31.8, 32.6, 40.2, 45.9, 55.9,
4.2, 71.3, 215.0. Anal. Calcd for C12H21NO: C, 73.80; H,
0.84; N, 7.17. Found: C, 74.02; H, 10.86; N, 7.02.
1
3
6
1
Meth yl Azela ld eh yd e (15). A solution methyl oleate
(24.68 g, 0.0832 mol) in 500 mL of a 2:1 mixture of MeOH/
CH Cl was cooled to -78 °C. Ozone was bubbled through the
solution until a persistent purple color appeared. The solution
was purged with N while being warmed to room temperature
2
2
Exp er im en ta l Section
2
Gen er a l Meth od s. All manipulations involving air-sensi-
tive reagents were carried out under house nitrogen, using
flame-dried glassware. Flash chromatography was carried out
using Merck 60 230-400 mesh silica gel as the stationary
phase. Thin-layer chromatography was performed with Merck
silica gel 60 F-254 glass plates (0.25 mm). Diethyl ether and
until all the color had dissipated. The ozonide was decomposed
by adding glacial acetic acid (35 mL) in one portion, followed
by powdered Zn (13 g, 0.20 mol) in small portions to control
heat evolution. After all the Zn was added, the mixture was
allowed to stir 30 min and then filtered through Celite to
2
remove unreacted Zn, and H O (150 mL) was added to the

5
010 J . Org. Chem., Vol. 63, No. 15, 1998
Scott et al.
mixture to prevent acetal formation. The solution was con-
centrated to about one-half initial volume in vacuo and added
by adding 15% NaOH (50 mL). The mixture was warmed to
room temperature with stirring over 45 min. The layers were
to saturated NaHCO
with CH Cl
and concentrated. The crude product was distilled to produce
3
(200 mL). The mixture was extracted
(3 × 250 mL). The organic phases were dried
2 2
separated, and the aqueous layer was extracted with CH Cl
2
2
(2 × 40 mL). The combined organic phases were dried,
concentrated, and purified by flash chromatography (ethyl
acetate/hexanes 10/90) to produce 4.37 g (80%) of 19 as a
colorless oil. On a larger scale (20.00 g), this procedure
produced an 86% yield of purified aldehyde (the reaction time
nonanal (bp 41 °C, 0.6 Torr) and 14.82 g (96%) of 15 (bp 93
°
1
1
C, 0.6 Torr) as a colorless oil. IR: 2930, 2860, 2720, 1720,
740, 1200, 1170 cm-1
.
1
H NMR (400 MHz): δ 1.31 (br s, 6),
-
1
1
.61 (br s, 4), 2.30 (t, 2, J ) 7.5), 2.41 (dt, 2, J ) 1.7, 7.3), 3.66
was extended to 2 h). IR: 2260, 1730 cm
.
H NMR (400
1
3
(
s, 3), 9.75 (t, 1, J ) 1.7). C NMR (125 MHz): δ 21.91, 24.76,
8.82, 28.87, 28.90, 33.94, 43.77, 51.36, 174.11, 202.62. Anal.
Calcd for C10 : C, 64.49; H, 9.74. Found: C, 64.50; H,
.78.
Meth yl 10-Cya n o-8-(h yd r oxym eth yl)d eca n oa te (17).
To a solution of pyrrolidine (0.3799 g, 5.342 mmol) and
anhydrous K CO (0.455 g) in ether (2 mL) at 0 °C was added
MHz): δ 0.02 (s, 6), 0.86 (s, 9), 1.28 (br s, 8), 1.53-1.77 (m, 5),
2
2.36-2.43 (m, 4), 3.44 (dd, 1, J ) 5.8, 10.2), 3.55 (dd, 1, J )
1
3
H
18
O
3
4.2, 10.2), 9.74 (t, 1, J ) 1.8). C NMR (100 MHz): δ -5.61,
-5.56, 15.01, 18.14, 21.97, 25.80, 26.57, 27.60, 28.97, 29.50,
30.50, 39.31, 43.76, 64.75, 120.05, 202.56. Anal. Calcd for
9
18 2
C H35NO Si: C, 66.40; H, 10.84; N, 4.30. Found: C, 66.64;
H, 10.74; N, 4.11.
2
3
dropwise from an addition funnel aldehyde 15 (0.9936 g, 5.335
mmol) in ether (2 mL). After addition was complete, the ice
bath was removed and the slurry stirred 5 h. The solution
12-Cya n o-10-[[[(1,1-d im eth yleth yl)d im eth ylsilyl]oxy]-
m eth yl]-3-h yd r oxy-2-m eth yld od eca n oic Acid (20). A so-
lution of LDA was prepared by adding n-butyllithium (3.28
mL, 8.59 mmol) to diisopropylamine (0.8697 g, 8.595 mmol)
in THF (12.5 mL) at 0 °C. A solution of propionic acid (0.3180
g, 4.293 mmol) in THF (7.5 mL) was added dropwise, and the
reaction mixture was allowed to stir at room temperature for
1 h. The enolate solution was cooled to 0 °C, and a solution of
aldehyde 19 (1.40 g, 4.30 mmol) in THF (10 mL) was added
dropwise. The reaction mixture was stirred at 0 °C for 1 h
and then poured into 1 N HCl (30 mL). The mixture was
concentrated in vacuo to remove most of the THF, and then
was filtered through a fine glass frit to remove K
rinsed with 5 mL of ether. The crude enamine was concen-
trated and added with CH CN (5 mL) to a reaction tube
containing acrylonitrile (0.3402 g, 6.411 mmol) in CH CN (5
mL). The tube was sealed and heated at approximately 80
C for 20 h. After the mixture was cooled to room temperature,
O was added (5 mL) and heating resumed for 1 h. The
mixture was poured into 5% aqueous HCl (20 mL) and
extracted with CH Cl
2 3
CO and
3
3
°
H
2
2
2
(3 × 30 mL). The organic extracts were
dried and concentrated. The product was purified by flash
the aqueous phase was extracted with CH
2
Cl
2
(3 × 40 mL).
chromatography (ethyl acetate/hexane 20/80 f 30/70) to obtain
The crude acid was dried, concentrated, and then dissolved in
1 M NaOH (50 mL), which was extracted with ether (2 × 15
mL). The aqueous layer was brought to pH ) 1 with 12 M
0
.93 g (73%) of aldehyde 16 as a slightly yellow, semipure oil.
. H NMR (500 MHz): δ 1.24 (br s,
), 1.38-1.45 (m, 1), 1.49-1.54 (m, 2), 1.59-1.69 (m, 2), 1.89-
.96 (m, 1), 2.20 (t, 2, J ) 7.5), 2.24-2.42 (m, 3), 3.56 (s, 3),
.55 (d, 1, J ) 1.5). 13C NMR (125 MHz): δ 15.02, 23.83, 24.63,
6.40, 28.39, 28.68, 29.09, 33.82, 50.08, 51.32, 118.98, 173.93,
02.83.
-
1
1
IR: 2240, 1740, 1720 cm
6
1
9
2
2
HCl and extracted with CH
organic phases were dried and concentrated to produce 1.33 g
2
Cl
2
(3 × 40 mL). The combined
(77%) of a 1:1 mixture of diastereomers of 20 as a viscous,
-1
1
slightly yellow oil. IR: 3420, 3120, 2220, 1720 cm
. H NMR
(500 MHz): δ 0.02 (s, 12), 0.87 (s, 18), 1.18 (d, 3, J ) 7.1), 1.22
(d, 3, J ) 7.2), 1.28 (br s, 20), 1.40-1.77 (m, 10), 2.39 (t, 4, J
) 7.5), 2.51-2.57 (m, 2), 3.44 (dd, 2, J ) 6.0, 10.2), 3.57 (dd,
To a solution of 16 (0.5935 g, 2.480 mmol) in MeOH (10 mL)
at 0 °C was slowly added NaBH
min, the solution was poured onto 5% HCl (10 mL) and
extracted with CH Cl
4
(0.0482 g, 1.274 mmol). After
1
3
5
2, J ) 4.2, 10.2), 3.65-3.69 (m, 1), 3.91-3.95 (m, 1). C NMR
(100 MHz): δ -5.59, -5.54, 10.36, 14.10, 15.02, 18.15, 25.33,
25.81, 26.71, 27.62, 29.31, 26.69, 30.55, 33.58, 34.47, 39.36,
44.11, 45.13, 64.89, 71.66, 73.15, 120.06, 180.58, 180.70. Anal.
2
2
(3 × 10 mL). The organic phase was
dried and concentrated, and the crude oil was purified by
Kugelrohr distillation to provide 0.5971 g (100%) of alcohol
-
1
1
1
7 as a colorless oil. IR: 3500, 2260, 1740 cm
.
H NMR
4
Calcd for C21H41NO Si: C, 63.11; H, 10.34; N, 3.50. Found:
(
500 MHz): δ 1.27 (br s, 8), 1.54-1.59 (m, 3), 1.61-1.68 (m,
C, 63.39; H, 10.40; N, 3.27.
1
2
1
2
1
), 1.70-1.77 (m, 1), 2.10 (br s, 1), 2.26 (t, 2, J ) 7.5), 2.37-
.41 (m, 2), 3.47 (dd, 1, J ) 6.0, 10.9), 3.58 (dd, 1, J ) 4.4,
Met h yl 12-Cya n o-10-[[[(1,1-d im et h ylet h yl)d im et h yl-
silyl]oxy]m eth yl]-3-h ydr oxy-2-m eth yldodecan oate (Meth -
yl Ester of 20. To acid 20 (5.42 g, 13.6 mmol) dissolved in
ether (50 mL) was added an excess of freshly prepared
diazomethane (from the reaction of Diazald and KOH) in ether.
The solution was concentrated to provide 5.61 g (100%) of the
ester as a 1:1 mixture of diastereomers as a slightly yellow
1
3
0.9), 3.62 (s, 3). C NMR (125 MHz): δ 14.90, 24.66, 26.41,
7.11, 28.80, 29.27, 30.32, 33.86, 39.26, 51.36, 64.27, 120.00,
3
74.25. Anal. Calcd for C13H23NO : C, 64.70; H, 9.61; N, 5.80.
Found: C, 64.44; H, 9.75; N, 5.98.
Met h yl 10-Cya n o-8-[[[(1,1-d im et h ylet h yl)d im et h yl-
-
1
1
silyl]oxy]m eth yl]d eca n oa te (18). To a solution of alcohol
1
oil. IR: 3500, 2240, 1740 cm
.
H NMR (500 MHz): δ 0.02
7 (11.02 g, 0.0457 mol) in DMF (20 mL) was added imidazole
(s, 12), 0.87 (s, 18), 1.16 (d, 3, J ) 7.2), 1.19 (d, 3, J ) 7.2),
1.27 (br s, 20), 1.34-1.76 (m, 10), 2.39 (t, 4, J ) 7.5), 2.48-
2.56 (m, 4), 3.43 (dd, 2, J ) 6.0, 10.2), 3.56 (dd, 2, J ) 4.1,
10.2), 3.62-3.65 (m, 1), 3.69 (s, 6), 3.83-3.87 (m, 1). C NMR
(100 MHz): δ -5.59, -5.54, 10.61, 14.25, 15.03, 18.16, 25.43,
25.82, 25.90, 26.75, 27.64, 29.38, 29.41, 30.57, 33.76, 34.65,
39.37, 44.21, 45.17, 51.63, 51.70, 64.83, 71.66, 73.24, 120.09,
(7.83 g, 0.115 mol) followed by tert-butyldimethylsilyl chloride
(8.26 g, 0.055 mol). The solution was stirred overnight at room
1
3
temperature and then poured onto 1 N HCl (200 mL) and
extracted with ether (2 × 225 mL). The combined organic
phases were then washed with 100-mL portions of 1 N HCl,
H
2
O, and saturated NaCl solution. The organic phase was
dried, concentrated, and purified by Kugelrohr distillation (bp
4
176.37. Anal. Calcd for C22H43NO Si: C, 63.87; H, 10.48; N,
2
25 °C at 0.3 Torr) to provide 15.43 g (95%) of the protected
3.39. Found: C, 64.05; H, 10.58; N, 3.59.
-1
1
alcohol 18. IR: 2240, 1740 cm
.
H NMR (500 MHz): δ 0.02
Met h yl 13-Am in o-10-[[[(1,1-d im et h ylet h yl)d im et h yl-
silyl]oxy]m eth yl]-3-h yd r oxy-2-m eth yltr id eca n oa te (21).
The methyl ester of 20 (5.17 g, 12.5 mmol) was dissolved in
ethyl acetate (200 mL) in a Parr bottle (500 mL capacity). To
the solution was added acetic acid (6.61 g, 110 mmol), followed
(
s, 6), 0.86 (s, 9), 1.27 (br s, 8), 1.55-1.76 (m, 5), 2.28 (t, 2, J
)
7.5), 2.38 (t, 2, J ) 7.5), 3.43 (dd, 1, J ) 6.0, 10.2), 3.55 (dd,
13
1
, J ) 4.2, 10.2), 3.64 (s, 3). C NMR (125 MHz): δ -5.63,
5.58, 14.99, 18.12, 24.79, 25.78, 26.58, 27.60, 28.94, 29.40,
0.50, 33.95, 39.31, 51.34, 64.75, 120.06, 174.10. Anal. Calcd
for C19 Si: C, 64.17; H, 10.49; N, 3.94. Found: C, 63.99;
H, 10.42; N, 3.99.
-[[[(1,1-Dim eth yleth yl)d im eth ylsilyl]oxy]m eth yl]-11-
oxou n d eca n itr ile (19). To a solution of ester 18 (6.00 g, 16.9
mmol) in CH Cl (60 mL) at -93 °C (MeOH/liquid N ) was
added DIBALH (18.5 mL, 1.0 M solution in CH Cl ) over 12
min. After being stirred for 8 min, the reaction was quenched
-
3
by PtO
an H atmosphere (53 psi) for 42 h. The solution was filtered
2
(1.01 g, 4.30 mmol). The solution was shaken under
H
37NO
3
2
through Celite and then extracted with 1 M NaOH (250 mL).
4
The aqueous layer was extracted with ethyl acetate (2 × 150
2 3
mL). The combined organic phases were dried (K CO ) and
2
2
2
concentrated to provide 5.06 g (97%) of the crude amine as a
colorless oil. IR: 3360, 1740, 1600 cm
MHz): δ 0.01 (s, 12), 0.87 (s, 18), 1.16 (d, 3, J ) 7.3), 1.18 (d,
-
1
1
2
2
.
H NMR (400

Synthesis of Petrosin, Petrosin A, and Petrosin B
J . Org. Chem., Vol. 63, No. 15, 1998 5011
3
3
3
, J ) 7.3), 1.20-1.50 (m, 34), 2.47-2.54 (m, 2), 2.65 (br s, 4),
.43 (dd, 2, J ) 5.7, 10.0), 3.46 (dd, 2, J ) 5.5, 10.0), 3.61-
23 (0.73 g, 1.1 mmol) was added to a 50 mL flask with MeOH
(7.5 mL), THF (10 mL), and 1 N NaOH (7.5 mL) and the
solution stirred at room temperature for 1.75 h. The solution
was reduced to approximately one-half volume under reduced
pressure followed by the addition of 1 N HCl (7.5 mL) until
pH ) 1. The acidic solution was extracted (EtOAc, 1 × 40
mL, 2 × 25 mL), and the combined organic layers were dried
and concentrated. The crude acid was added to a 25 mL flask
with the aid of THF (8 mL) and cooled in an ice-water bath.
Pentafluorophenol (0.59 g, 3.2 mmol) was added in THF (2
mL), followed by the addition of dicyclohexylcarbodiimide (0.26
g, 1.3 mmol), and the solution stirred for 3 h. The cold bath
was removed and the solution allowed to stir at room temper-
ature for 10 h. The solution was filtered through a glass wool
plug to remove the solid dicyclohexylurea formed and concen-
trated. To the crude active ester was added dioxane saturated
with anhydrous HCl (16 mL) and the solution stirred for 1 h
at room temperature. The solution was concentrated to an
oil and was taken up into a gastight syringe along with dioxane
(5 mL) and THF (5 mL). The flask was rinsed (3 mL of
dioxane, 5 mL of THF) and the rinse combined in the syringe.
The solution was added via syringe pump over 3.5 h to a 5:1
dioxane/pyridine solution (600 mL) maintained at 85-90 °C.
The syringe was rinsed with THF (3 mL) and the rinse added
at the same rate. The solution was stirred in the heated bath
for an additional 4.5 h and then was cooled and concentrated.
The crude oil was purified by flash chromatography (1:9
1
3
.66 (m, 1), 3.69 (s, 6), 3.83-3.86 (m, 1).
C NMR (100
MHz): δ -5.42, 10.70, 14.22, 18.26, 25.45, 25.90, 26.70, 28.15,
2
5
C H47NO Si: C, 63.26; H, 11.34; N, 3.35. Found: C, 63.40;
22 4
H, 11.27; N, 3.57.
Meth yl 13-[[12-Cyan o-10-[[[(1,1-dim eth yleth yl)dim eth yl-
silyl]oxy]m e t h yl]-3-h yd r oxy-2-m e t h yl-1-oxod od e cyl]-
a m in o ]-10-[[[(1,1-d im e t h y le t h y l)d im e t h y ls ily l]o x y ]-
m eth yl]-3-h yd r oxy-2-m eth yltr id eca n oa te (22). Acid 20
9.45, 29.91, 30.85, 33.92, 34.66, 40.26, 42.64, 44.33, 45.29,
1.61, 51.69, 65.55, 71.63, 73.17, 176.40. Anal. Calcd for
(2.98 g, 7.46 mmol) and amine 21 (3.11 g, 7.45 mmol) were
added to a 50 mL flask with THF (23 mL). To the solution
was added hydroxybenzotriazole hydrate (1.20 g, 8.91 mmol),
and after dissolution the solution was cooled in an ice-water
bath. Dicyclohexylcarbodiimide (1.84 g, 8.90 mmol) was added,
and the solution was stirred in the ice bath for 1 h. The cold
bath was removed and the solution allowed to warm to room
temperature. After 12 h, the solution was filtered to remove
the solid dicyclohexylurea. After concentration, the crude
product was purified by flash chromatography to provide 4.20
g (70%) of amide 22. IR: 3350, 2250, 1740, 1645, 1545, 1465
-
1
1
17
cm
.
H NMR (500 MHz): δ 0.01 (s, 6), 0.02 (s, 6), 0.87 (s,
1
2
3
8), 1.13-1.76 (m, 38), 2.17-2.26 (m, 1), 2.40 (t, 2, J ) 7.5),
.49-2.55 (m, 1), 3.17-3.25 (m, 2), 3.40-3.49 (m, 4), 3.55-
.58 (m, 2), 3.60-3.67 (m, 1), 3.69 (s, 3), 3.80-3.88 (m, 1), 5.90
1
3
17
(br s, 1). C NMR (100 MHz): δ -5.57, -5.52, -5.43, -5.42,
2 2 2 2
MeOH/CH Cl f 1.5:8.5 MeOH/CH Cl ) to provide 0.46 g (78%)
of the macrocycle 24 as a slightly yellow glass. The absence
1
2
2
3
4
1
0.69, 11.10, 14.24, 15.06, 15.82, 18.17, 18.26, 25.41, 25.75,
5.83, 25.91, 25.96, 26.66, 26.74, 26.84, 26.88, 27.64, 28.33,
9.39, 29.43, 29.74, 29.76, 29.82, 30.57, 30.83, 33.55, 33.83,
4.67, 35.45, 39.36, 39.61, 39.65, 40.06, 44.28, 44.56, 45.21,
5.90, 51.64, 51.71, 64.92, 65.52, 71.74, 71.91, 73.28, 73.75,
of dimeric products was confirmed by FAB-MS. IR (KBr
-
1
1
pellet): 3320, 1640, 1540, 1460 cm
3
. H NMR (500 MHz, CD -
OD):17 δ 1.16 (d, 6, J ) 6.8), 1.20-1.57 (m, 34), 2.18-2.26 (m,
1
3
1), 2.35-2.43 (m, 1), 2.79-3.62 (m, 10). C NMR (100 MHz,
CD
3
OD):17 δ 14.98, 15.03, 15.14, 15.20, 26.79, 26.98, 27.70,
20.13, 175.98, 176.41, 176.53. Anal. Calcd for C43
: C, 64.61; H, 10.85; N, 3.51. Found: C, 64.76; H, 10.84;
N, 3.64.
H
86
N
2
O
7
-
Si
2
27.86, 29.30, 30.70, 31.00, 31.13, 31.21, 32.01, 32.05, 32.13,
32.20, 36.17, 36.25, 40.42, 40.50, 41.46, 41.50, 41.57, 41.62,
4
1
5
6.55, 46.62, 48.87, 65.56, 65.63, 74.09, 74.63, 74.74, 177.91,
77.99. Anal. Calcd for C30 : C, 66.38; H, 10.77; N,
.15. Found: C, 66.03; H, 10.44; N, 5.04.
Meth yl 13-[[13-[[(1,1-Dim eth yleth oxy)car bon yl]am in o]-
-h yd r oxy-10-(h yd r oxym eth yl)-2-m eth yl-1-oxotr id ecyl]-
58 2 6
H N O
3
a m in o]-3-h yd r oxy-10-(h yd r oxym et h yl)-2-m et h ylt r id e-
ca n oa te (23). The amide 22 (7.78 g, 9.73 mmol) was added
to a Parr bottle (500 mL capacity) with 0.23 M HCl in ethanol
solvent (4 mL of concd HCl + 200 mL of absolute ethanol),
followed by PtO
under an H atmosphere (53 psi) for 18 h. The solution was
filtered through Celite and concentrated. To the crude oil were
added H O (20 mL) and dioxane (20 mL), followed by the slow
addition of NaHCO (10 g). A solution of di-tert-butyl dicar-
Bis(1,1-d im et h ylet h yl) 4,18-Dih yd r oxy-11,25-b is(h y-
d r oxym eth yl)-3,17-d im eth yl-1,15-d ia za cycloocta cosa n e-
,15-d ica r boxoa te (26). The bis-amide 24 (0.2813 g, 5.183
mmol) was added to a 50 mL flask with the aid of methanol
approximately 10 mL). The solution was reduced in volume
with a rotary evaporator until an oil remained, and THF (35
mL) was added. Solid LiAlH (0.6260 g, 16.50 mmol) was
1
2
(0.23 g, 1.0 mmol). The solution was shaken
(
2
4
2
added in portions slowly. The solution was heated at reflux
for 23 h. TLC of the solution showed that the starting material
3
bonate (2.56 g, 11.7 mmol) in dioxane (12 mL) was added, and
the mixture was allowed to stir for 12 h. The solution was
was still present, so additional LiAlH
was added and heating continued for 13 h. To the cooled
solution were added H O (0.8411 g), 15% aqueous NaOH
0.8670 g), and H O (2.5201 g), and the solution was stirred a
few minutes until all the solids were white. The solution was
dried with Na SO and filtered through a glass frit, rinsing
4
(0.2090 g, 5.507 mmol)
concentrated and added to a separatory funnel with H
mL) and CH Cl (100 mL). The aqueous layer was extracted
further with CH Cl
(2 × 40 mL), and the combined organic
layers were dried and concentrated. The crude oil was purified
by flash chromatography (1:9 MeOH/CH Cl ) to provide 4.75
2
O (50
2
2
2
(
2
2
2
2
4
2
2
several times with small portions of THF. The solution was
concentrated to provide 0.2743 g (103%) of crude diamine 25.
The crude amine was added to a 10 mL flask with THF and
g (72%) of 23 as a colorless oil. IR: 3330, 1710, 1685, 1640,
1
530, 1450, 1360 cm-1
1
17
.
H NMR (500 MHz): δ 1.12-1.56
(m, 49), 2.20-2.60 (m, 4), 2.76-2.90 (m, 1), 3.02-3.35 (m, 4),
2
rotovapped down to an oil. Dioxane (1 mL) and H O (0.6 mL)
3
1
1
2
3
4
6
1
.40-3.67 (m, 5), 3.69 (s, 3), 3.73-3.90 (m, 2), 4.60-4.80 (m,
1
3
17
were added followed by di-tert-butyl carbonate (0.2717, 1.245
mmol) in dioxane (0.6 mL). Bubbling was evident immedi-
ately. The solution was stirred for 45 min and the solvent
removed. Purification by flash chromatography (1:9 MeOH/
), 6.34-6.45 (m, 1). C NMR (100 MHz): δ 10.89, 11.30,
3.89, 15.50, 15.54, 25.29, 25.42, 25.67, 25.75, 26.44, 26.62,
7.16, 27.88, 28.29, 29.18, 29.26, 29.59, 29.70, 30.67, 30.79,
3.50, 33.97, 34.34, 35.27, 39.18, 39.33, 39.77, 39.87, 39.95,
0.74, 44.52, 44.68, 45.33, 45.65, 45.84, 50.25, 51.51, 51.57,
4.80, 64.87, 71.73, 71.93, 73.06, 73.48, 73.55, 78.93, 156.16,
CH
2 2
Cl ) provided 0.3337 g (90%) of the di-Boc tetraol 26 as a
-
1
1
17
colorless glass. IR: 3420, 1670 cm
.
H NMR (500 MHz):
δ 0.78-0.82 (m, 3), 0.87-0.92 (m, 3), 1.18-1.90 (m, 54), 2.59-
76.28, 176.35, 176.61. Anal. Calcd for C36
70 2 9
H N O : C, 64.06;
4
2
2
3
3
6
1
.05 (m, 18). 13C NMR (100 MHz): δ 9.95, 10.15, 15.18, 25.49,
5.94, 25.99, 26.07, 26.16, 26.20, 26.36, 26.57, 27.40, 27.48,
7.73, 27.81, 28.35, 28.42, 29.17, 29.35, 29.52, 29.77, 30.12,
0.21, 30.54, 30.64, 31.12, 33.38, 34.27, 36.20, 36.53, 37.71,
8.20, 38.29, 39.88, 40.08, 48.37, 49.34, 49.66, 51.57, 52.17,
5.43, 65.65, 68.37, 68.65, 69.02, 74.25, 79.46, 79.56, 79.93,
17
H, 10.45; N, 4.15. Found: C, 63.70; H, 10.09; N, 3.97.
,18-Dih yd r oxy-11,25-b is(h yd r oxym et h yl)-3,17-d im e-
th yl-1,15-d ia za cycloocta cosa n e-2,16-d ion e (24). The ester
4
(17) Because of the large number of diastereomers present in
compounds 22-26, the spectra were often largely uninterpretable. For
these compounds, only the common multiplets are listed for H NMR.
This is also the reason for the very large number of peaks in the
NMR spectra. However, the data as presented are still useful for
structure determination.
56.29, 157.22. Anal. Calcd for C40
78 2 8
H N O : C, 67.19; H,
1
1
3
11.00; N, 3.92. Found: C, 66.90; H, 10.66; N, 3.97.
F in a l Rin g Closu r e. The tetraol 26 (0.2436 g, 0.3407
mmol) was dissolved in CH Cl (13 mL). To the stirred
2 2
C

5
012 J . Org. Chem., Vol. 63, No. 15, 1998
Scott et al.
solution was added solid Dess-Martin periodinane (1.0443 g,
with CH
2
Cl
2
(3 × 3 mL), and the solution concentrated. The
2
1
.462 mmol) and the solution stirred at room temperature for
h. The reaction mixture was added to a separatory funnel
crude oil was added to a reaction tube with propylammonium
acetate (0.0076 g, 0.0638 mmol) and 1,2-dichloroethane (2 mL)
and the solution heated at approximately 95 °C for 7 days. To
containing 1 M NaOH (40 mL) and Et
separation of the layers, the aqueous layer was extracted with
Et
O (1 × 40 mL). The combined organic layer was reduced
to an oil, and 95% ethanol (40 mL), H O (8 mL), and 12 M
2
O (40 mL). After
the solution was added H
1 h. The mixture was added to a separatory funnel with 1 M
NaOH (5 mL) and CH Cl (10 mL), and the layers were
separated. The aqueous layer was extracted with CH Cl (3
× 7 mL), and the combined organic layers were dried with
CO Concentration produced the crude oil, which was
purified by flash chromatography (9.5:9.5:1 hexane/Et O/Et N)
to produce 0.0245 g (80%) of a mixture of isomers enriched in
petrosin. To the crude oil was added Et O (1 mL) and the
2
O (1 mL) and heating resumed for
2
2
2
2
HCl (4 mL) were added. The solution was refluxed for 3 h.
The solution was concentrated to a thin oil and added to a
2
2
seperatory funnel with 1.25 M NaOH (40 mL) and CHCl
mL). The layers were separated, the aqueous layer was
extracted with CHCl
(2 × 20 mL), and the combined organic
layers were concentrated. To the crude mixture were added
5% ethanol (40 mL) and acetic acid (0.5108 g). The solution
3
(40
K
2
3
.
2
3
3
2
9
solution put in the freezer overnight to induce crystallization.
The mixture was centrifuged, and crystals of pure petrosin
(0.0080 g) were obtained along with an oily mixture of isomers
(0.0164 g).
was heated at reflux for 24 h. After concentration of the
solvent, the oil was added to a separatory funnel with 1 M
NaOH (30 mL) and CH
extracted with CH Cl
(2 × 20 mL), and the combined organic
layers were dried over K CO . Filtration and concentration
produced the crude oil, which was purified by flash chroma-
tography (9.5:9.5:1 hexane/Et O/Et N) to produce 0.0988 g
62%) of the petrosin skeleton as a mixture of stereoisomers.
2 2
Cl (30 mL). The aqueous layer was
2
2
2
3
Ack n ow led gm en t. We thank the National Insti-
tutes of Health for support of this research (GM 46057).
R.W.S. thanks the National Science Foundation and the
Division of Organic Chemistry of the American Chemi-
cal Society for graduate fellowships. We also thank
Professor Isao Kitagawa for a comparison sample of
natural petrosin.
2
3
(
2
The oil was dissolved in Et O (1.75 mL) and put in the freezer
overnight to induce crystallization. Centrifugation and de-
cantation, followed by washing of the crystals with ether and
repeating the process, provided 0.0374 g of pure petrosin (1)
and 0.0601 g of petrosin isomers. The petrosin was identical
1
13
to an authentic sample by H NMR, C NMR, IR, and TLC.
Su p p or tin g In for m a tion Ava ila ble: Separation scheme
for isolation of petrosin, petrosin A, petrosin B, and petrosin
B′, table of C NMR chemical shifts for the four isomers, and
calculated molecular mechanics energies for 24 isomers of
compound 4 (4 pages). This material is contained in libraries
on microfiche, immediately follows this article in the microfilm
version of the journal, and can be ordered from the ACS; see
any current masthead page for ordering information.
1
3
The isomers were identified mainly by C NMR, IR, and MS.
Equ ilibr a tion (Typ ica l P r oced u r e, En tr y 3 fr om Ta ble
13
3
P r ovid ed Below ). To one-half of the mixture of petrosin
isomers from above (after removal of crystalline petrosin,
.0305 g, 0.0648 mmol) were added butylamine (1.5 mL) and
0
several 3 Å molecular sieves. The solution was heated in a
sealed tube at approximately 95 °C until the ketone peak was
no longer clearly observed in the IR (68 h). The crude imine
was separated from the molecular sieves, which were rinsed
J O9801768